The maximum entropy model was introduced and a new intrusion detection approach based on the maximum entropy model was proposed. The vector space model was adopted for data presentation. The minimal entropy partitioni...The maximum entropy model was introduced and a new intrusion detection approach based on the maximum entropy model was proposed. The vector space model was adopted for data presentation. The minimal entropy partitioning method was utilized for attribute diseretization. Experiments on the KDD CUP 1999 standard data set were designed and the experimental results were shown. The receiver operating eharaeteristie（ROC） curve analysis approach was utilized to analyze the experimental results. The analysis results show that the proposed approach is comparable to those based on support vector maehine（SVM） and outperforms those based on C4.5 and Naive Bayes classifiers. According to the overall evaluation result, the proposed approach is a little better than those based on SVM.展开更多
篇章关系分为显式和隐式两种。显式关系的显著特征是篇章的基本单元之间存在显式连接词。针对汉语显式篇章关系,构建了包括汉语连接词识别和篇章关系分类的显式篇章关系分析平台。该文选取汉语宾州树库(Chinese Penn Treebank,CTB)中的...篇章关系分为显式和隐式两种。显式关系的显著特征是篇章的基本单元之间存在显式连接词。针对汉语显式篇章关系,构建了包括汉语连接词识别和篇章关系分类的显式篇章关系分析平台。该文选取汉语宾州树库(Chinese Penn Treebank,CTB)中的500篇文本进行了汉语显式篇章关系标注;结合连接词的中心词,采用最大熵分类器构建了汉语连接词识别模块,其性能F1值达到了66.79%;基于连接词及其词性等上下文特征,构建了篇章关系分类器,其在最顶层4大类语义关系上的分类性能的F1值为91.92%。展开更多
The performance of scene classification of satellite images strongly relies on the discriminative power of the low-level and mid-level feature representation. This paper presents a novel approach, named multi-level ma...The performance of scene classification of satellite images strongly relies on the discriminative power of the low-level and mid-level feature representation. This paper presents a novel approach, named multi-level max-margin analysis （M 3 DA） for semantic classification for high-resolution satellite images. In our M 3 DA model, the maximum entropy discrimination latent Dirichlet allocation （MedLDA） model is applied to learn the topic-level features first, and then based on a bag-of-words repre- sentation of low-level local image features, the large margin nearest neighbor （LMNN） classifier is used to optimize a multiple soft label composed of word-level features （generated by SVM classifier） and topic-level features. The categorization performances on 21-class land-use dataset have demonstrated that the proposed model in multi-level max-margin scheme can distinguish different categories of land-use scenes reasonably.展开更多
基金Supported bythe National Research Foundationforthe Doctoral Program of Higher Education of China(20030145029) the Teaching and Research Award Program for Outstanding Young Teachers in Higher Education Institutions of the Ministry ofEducation
文摘The maximum entropy model was introduced and a new intrusion detection approach based on the maximum entropy model was proposed. The vector space model was adopted for data presentation. The minimal entropy partitioning method was utilized for attribute diseretization. Experiments on the KDD CUP 1999 standard data set were designed and the experimental results were shown. The receiver operating eharaeteristie（ROC） curve analysis approach was utilized to analyze the experimental results. The analysis results show that the proposed approach is comparable to those based on support vector maehine（SVM） and outperforms those based on C4.5 and Naive Bayes classifiers. According to the overall evaluation result, the proposed approach is a little better than those based on SVM.
文摘篇章关系分为显式和隐式两种。显式关系的显著特征是篇章的基本单元之间存在显式连接词。针对汉语显式篇章关系,构建了包括汉语连接词识别和篇章关系分类的显式篇章关系分析平台。该文选取汉语宾州树库(Chinese Penn Treebank,CTB)中的500篇文本进行了汉语显式篇章关系标注;结合连接词的中心词,采用最大熵分类器构建了汉语连接词识别模块,其性能F1值达到了66.79%;基于连接词及其词性等上下文特征,构建了篇章关系分类器,其在最顶层4大类语义关系上的分类性能的F1值为91.92%。
基金Supported by the Open Projects Program of National Laboratory of Pattern Recognition and the National Natural Science Foundation of China(91338113)
文摘The performance of scene classification of satellite images strongly relies on the discriminative power of the low-level and mid-level feature representation. This paper presents a novel approach, named multi-level max-margin analysis （M 3 DA） for semantic classification for high-resolution satellite images. In our M 3 DA model, the maximum entropy discrimination latent Dirichlet allocation （MedLDA） model is applied to learn the topic-level features first, and then based on a bag-of-words repre- sentation of low-level local image features, the large margin nearest neighbor （LMNN） classifier is used to optimize a multiple soft label composed of word-level features （generated by SVM classifier） and topic-level features. The categorization performances on 21-class land-use dataset have demonstrated that the proposed model in multi-level max-margin scheme can distinguish different categories of land-use scenes reasonably.
